This project supports the continuation of the PI1's interdisciplinary training started in September 1999 with a Sloan/DOE fellowship in Computational Molecular Biology. The PI1's immediate career objective is to complete a transition into computational molecular biology by forging an independent research trajectory within this exciting field. University of Washington is a uniquely suitable place for making such a transition: it has several strong biology programs as well as a strong computer science department, with considerable synergy among them through active collaborations, joint courses, and seminars. The PI its a computer scientist by training, and based on initial investigations is convinced that her expertise in designing efficient algorithms can be gainfully employed to address challenging computational problems arising from the emergent need for global analysis of massive biological datasets, for example, the human genome. Her long term interest in this field, however, is fueled largely by the tantalizing vision of contributing to real biological knowledge. To realize this vision, she plans to acquire expertise in experimental techniques that will allow her to test biological hypotheses arising from computational analyses, through an extensive regimen of hands-on, lab-intensive coursework. The specific research problem the FLPI proposes to explore is to apply information retrieval techniques (which have been widely applied with great success in developing search engines for the world-wide web) for large-scale analysis of genomic sequence and gene expression data. She plans to build computational models of the untranslated promoter regions upstream of coexpressed eukaryotic genes, by identifying composite regulatory motifs that determine the genes' specific expression pattern. -Such composite motifs may be composed of the binding sites of several regulatory factors that coordinately control expression. These models are then to be used to classify promoter regions of unannotated genes, thereby providing a hypothesis for their function. This research has the potential to lead to novel computational methods for classification of eukaryotic promoter regions in particular, and for functional genomics in general. The intensive training afforded through the planned training activities and the proposed research will enable the PI1 to successfully complete the transition into being an independent investigator in computational molecular biology.